Enhancing National Flood Forecasting: Leveraging Library-based Surrogate Models for Impact-based Warnings

In Switzerland, the Federal Office for the Environment issues hydrological forecasts and general flood warnings for the main river network. However, recent global flood events underscore that gaps in the communication channels from warning services to target groups inhibit effective mitigation efforts. One approach addressing this issue is impact-based warning.

Aligned with Switzerland's existing flood forecasting system, we introduce a library-based surrogate flood model approach aimed at advancing current technologies towards robust impact-based warning systems. We evaluate the model based on the main river network of Northern Switzerland by comparing the impacts to buildings, persons and workplaces with hazard classification, estimated with transient simulations for nine extreme precipitation scenarios.

Across 78 analyzed model regions, our surrogate approach yields a Flood Area Index between 0.74 and 0.90 for each scenario (overall 0.84) compared to the transient, computationally expensive flood modelling approach. Furthermore, the Critical Success Index, computed based on exposed persons, ranges between 0.77 and 0.93 (overall 0.89).

Our prototype of a library-based flood surrogate model demonstrates the ability of accurately replicate highly resolved transient models This capability bears the potential of nationwide real-time flood impact prediction and potential integration into probabilistic forecasting. Leveraging an API, this library-based approach could enhance the existing forecasting system, offering a pathway toward impact-based flood warnings.